Supplementary Materials: Improving Deep Learning Interpretability by Saliency Guided Training

We compare interpretable and traditional training for different saliency methods with modification-based evaluation. Each experiment is repeated five times.